Biology subjects

Verma, A.

Publications and source records attributed to Verma, A..

At least 19 recordsLinked to original sources

bioRxiv2026

In Silico Design and Evaluation of CRISPR Cytosine Base Editing Strategy for Potential Correction of the ASPA Variant Associated with Canavan Disease

**Background** Canavan disease is a rare autosomal recessive neurodegenerative leukodystrophy caused by pathogenic variants in the ASPA gene, for which effective molecularly targeted therapies remain limited. The NM_000049.4:c.691T>C (p.Tyr231His) variant introduces a pathogenic C-G base pair that is potentially amenable to correction using cytosine base editing (CBE). This study aimed to design and computationally evaluate a CBE-based CRISPR strategy for potential correction of this ASPA variant. **Methods** An integrated in silico workflow was used to characterize the ASPA p.Tyr231His variant and design a CBE-compatible sgRNA. Variant pathogenicity, conservation, and structural stability were assessed using multiple computational tools. Candidate sgRNAs were designed and evaluated for editing-window positioning, PAM compatibility, specificity, GC content, bystander editing, and secondary structure. Genome-wide off-target potential was further assessed using CRISPRitz against the GRCh38 reference genome. **Results** The ASPA p.Tyr231His variant was exceptionally rare (gnomAD allele frequency: 1.86 x 10^-6), highly conserved, and consistently predicted to be deleterious, with DynaMut2 indicating reduced protein stability (DDG = -2.04 kcal/mol). Of two candidate sgRNAs, one was selected based on optimal editing-window positioning, absence of predicted bystander editing, favourable GC content (45%), and specificity scores (MIT: 82; CFD: 93). RNAfold predicted minimal secondary-structure constraints (MFE: 0.00 kcal/mol). CRISPRitz identified 108 potential off-target sites, with none containing fewer than three mismatches, no bulge-associated sites, and only three predicted sites within exonic regions. **Conclusions** The integrated computational analysis identified a candidate sgRNA with favourable predicted editing characteristics and genome-wide specificity for potential CBE-mediated correction of the ASPA c.691T>C (p.Tyr231His) variant. These findings provide a rational framework for prioritizing this guide for experimental evaluation. Further cell-based and in vivo studies are required to establish editing efficiency, editing purity, bystander activity, off-target effects, and restoration of ASPA function before therapeutic applicability can be assessed.

bioinformatics↗

bioRxiv2026

Benchmarking Large Language Models for Assay-Aware Drug-Target Binding Affinity Prediction

Machine learning models for molecular discovery often train on bioactivity data from heterogeneous assays, yet assay context is rarely explicitly represented. We investigate whether natural-language assay descriptions provide predictive information beyond molecular and protein representations. We develop ACL-DTI (Assay-aware Contrastive Learning-based Drug Target Interaction) and MAMMAL-AsAP, an assay-context extension of MAMMAL-DTI. Across five triplet-disjoint random splits, assay-aware models achieve competitive or improved performance over assay-agnostic baselines. Among the evaluated models, MAMMAL-AsAP achieves the highest (R^2) of 0.7198 +/- 0.0085. Masking experiments show that ACL-DTI is robust to incomplete assay descriptions, exhibiting the smallest reductions in R^2 among the evaluated models when masking the target name, assay technique, or both: 0.0228 +/- 0.0064, 0.0406 +/- 0.0113, and 0.0620 +/- 0.0190, respectively. These results indicate that assay descriptions contain predictive information not fully captured by molecular structures and protein sequences, and reveal architecture-dependent differences in how this information is exploited.

bioinformatics↗

bioRxiv2026

ARCHER-LD: Rapid Long-Range Linkage Disequilibrium Calculations at Biobank Scale using GPU Acceleration

Linkage disequilibrium (LD) information from one's own dataset is considered optimal for downstream analyses such as statistical fine-mapping. However, the computational complexity ({approx}N2 / 2 computations for N variants) lead most studies to use external reference panels, such as from 1000 Genomes. To capture LD across all variant pairs in biobank-scale whole-genome sequencing (WGS) datasets with hundreds of millions of variants, new computational strategies are essential. We present a novel approach that uses multi-GPU distributed computing to compute R2 for every variant pair in a dataset. On chromosome 22 (1.8 million variants) of the 30x WGS 1000 Genomes dataset, our method using eight consumer-level 12GB GPUs (NVIDIA RTX 2080TIs) takes <10 minutes, while the same calculation with PLINK using a high-end 64-threaded CPU (Intel Xeon Gold 6338) takes >80 minutes, corresponding to a {approx}8x speedup. We further demonstrate true biobank-scale performance in the Penn Medicine Biobank (PMBB; 57,170 samples), computing chromosome 1 LD ({approx}1.38 million variants) in 49 minutes versus 22.7 hours for PLINK, a {approx}28x speedup. With this method, we successfully computed on the aforementioned 30x WGS 1000 Genomes dataset ({approx}120 million variants and {approx}2500 samples) the entire LD matrix (>1e16, or 10 quadrillion elements) in under 6 hours using 512 NVIDIA 40GB A100 GPUs on the Department of Energy Argonne Leadership Computing Facility Polaris Supercomputer. We make this tool, coded in Python using CuPy, publicly available. Using this tool, researchers can leverage the full extent of their genomic data without relying on external LD reference panels and acquire more accurate, population-specific findings, particularly for groups underrepresented in existing databases.

bioinformatics↗

bioRxiv2026

A Comparative Benchmark of Biomedical Language Models for Concept Normalization from Real-World Text

Patient-reported and clinically documented narratives often contain informal, fragmented and linguistically heterogeneous expressions, complicating medical concept normalization (MCN) and frequently necessitating pre-processing before terminology mapping. Despite rapid advances in biomedical language modeling, the comparative utility of representation models for MCN and their integration with instruction-tuned LLMs in automated normalization workflows remain underexplored. In this study, we systematically benchmarked 15 general, biomedical and clinical transformer-based representation models together with 12 instruction-tuned LLMs across 12,713 instances from five established datasets: TAC2017_ADR, TwADR-L, TwiMed, CADEC and SMM4H2017. Representation models were evaluated using embedding-based semantic retrieval, whereas instruction-tuned LLMs were assessed as upstream text-correction modules. SapBERT achieved the highest Top-5 accuracy among representation models, reaching 63.8% for SNOMED CT and 58.0% for MedDRA without correction. Qwen 2 Instruct was selected as the preferred corrector on the basis of its favorable balance between Top-1 normalization performance and computational efficiency relative to substantially larger models, including Llama 3.1 Instruct (70B). Incorporation of Qwen 2 Instruct increased Top-5 accuracy to 69.4% for SNOMED CT and 63.2% for MedDRA. The resulting framework accepts heterogeneous short medical expressions without manual input pre-processing and automatically performs text refinement, semantic retrieval and terminology mapping to standardized concepts and vocabulary codes. These findings establish a benchmark-guided, scalable framework for automated medical terminology standardization.

bioinformatics↗

bioRxiv2026

LINE-1 transposon derepression and epigenetic remodeling in Retinoblastoma

Retinoblastoma (RB), the most common pediatric intraocular malignancy, is initiated by biallelic inactivation of the RB1 tumor suppressor gene. Although LINE-1 (L1) retrotransposon activation has been linked to genomic instability and tumor evolution in several adult cancers, its contribution to pediatric malignancies remains poorly understood. Here, we investigated L1 expression in retinoblastoma samples from an Indian pediatric cohort by integrating transcriptomic, locus-specific, and immunohistochemistry. Transcriptomic profiling revealed increased L1 expression in RB tumors compared with control retinal tissues, which was validated by ORF1p immunostaining in tumor sections. Locus-specific analysis identified a subset of transcriptionally active L1 loci, suggesting selective activation of young L1 elements. We also observed altered expression of genes associated with proliferation, chromatin regulation, and oncogenic signaling, including MYCN, MDM2, E2F3, SOX4 and CREBBP. Our findings identify previously underexplored L1 activation as a molecular feature associated with retinoblastoma and suggest that disruption of transposon silencing accompanies the RB transcriptional landscape. Validation in larger patient cohorts will be important; however, this study provides a foundation for exploring L1 activity as a potential molecular marker and future therapeutic vulnerability in retinoblastoma. HighlightsO_LITranscriptomic profiling reveals increased LINE-1 expression in retinoblastoma. C_LIO_LIORF1p protein is robustly expressed in retinoblastoma. C_LIO_LIA subset of LINE-1 loci exhibits transcriptional activation, suggesting locus-specific derepression. C_LI In BriefRetinoblastoma is initiated by RB1 loss, but the contribution of transposable elements to its molecular landscape remains poorly understood. Here, we show that retinoblastoma is associated with LINE-1 promoter hypomethylation, increased LINE-1 transcript expression, ORF1p accumulation, and preferential activation of selected LINE-1 loci. These changes occur together with altered expression of oncogenic signalling pathways and host factors involved in transposable element repression, identifying dysregulated LINE-1 control as a molecular feature of this paediatric cancer. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=115 SRC="FIGDIR/small/740688v1_ufig1.gif" ALT="Figure 1"> View larger version (24K): org.highwire.dtl.DTLVardef@197e89org.highwire.dtl.DTLVardef@1c26c04org.highwire.dtl.DTLVardef@14180a4org.highwire.dtl.DTLVardef@16493bb_HPS_FORMAT_FIGEXP M_FIG C_FIG Retinoblastoma exhibits epigenetic activation of LINE-1 characterized by promoter hypomethylation, increased expression of young LINE-1 elements, and ORF1p accumulation. This is accompanied by coordinated transcriptomic remodeling and disruption of host mechanisms that normally restrain transposable elements, linking LINE-1 activation to the molecular landscape of retinoblastoma.

cancer biology↗

bioRxiv2026

PMBB Geno-Pheno Toolkit: A suite of scalable, reproducible pipelines for cross-biobank association analyses

SummaryElectronic health record (EHR)-linked biobanks generate unprecedented genomic and phenotypic datasets, but their scientific utility is constrained by data fragmentation across institutional silos and incompatible computing infrastructures, forcing researchers to rewrite ad-hoc scripts for each new environment. We present the PMBB Geno-Pheno Toolkit, a suite of modular Nextflow pipelines for biobank-scale association analyses. This note focuses on the toolkits SAIGE family of pipelines -- supporting genome-wide (GWAS), exome-wide (ExWAS), and phenome-wide (PheWAS) association testing -- together with the companion GWAMA and ExWAS meta-analysis pipelines that enable cross-biobank replication. All components are containerized (Docker/Apptainer) and orchestrated with Nextflow, allowing the same workflows to run unmodified on local HPC clusters, cloud platforms, and the All of Us Research Workbench. Complementary toolkit pipelines for PLINK-based GWAS, polygenic scoring, LD-based clumping, and phenotype harmonization are also available and briefly noted. AvailabilityThe PMBB Geno-Pheno Toolkit is freely available at https://github.com/PMBB-Informatics-and-Genomics/pmbb-geno-pheno-toolkit under MIT open-source license.

bioinformatics↗

bioRxiv2026

AI-Informed neoantigen prioritization enables a multi-epitope mRNA/LNP vaccine with antigen-specific immunogenicity and antitumor activity

Personalized neoantigen vaccines are an emerging strategy for cancer immunotherapy, but their effectiveness depends on selecting tumor-specific antigens capable of inducing functional T-cell responses. The VACINUS AI-informed neoantigen prioritization framework previously identified and peptide-validated three immunogenic Tier 1 neoantigens in the B16F10 melanoma model. In this study, we extended that framework by translating these validated neoantigens into a multi-epitope messenger RNA vaccine formulated with lipid nanoparticles and evaluating its preclinical immunogenicity and antitumor activity. The three VACINUS-prioritized B16F10 neoantigens were encoded within a single multi-epitope construct, BF-V1_27-Ser, and formulated to generate BF-RNA-P. In B16F10 tumor-bearing mice, BF-RNA-P induced neoantigen-specific CD8+T-cell responses, with the strongest response directed against the B16F10-1-4 epitope. Combination with anti-PD-1 further enhanced vaccine-induced CD44+IFN-{gamma}+ CD8+ T-cell activation, whereas anti-PD-1 alone did not induce detectable peptide-specific responses. BF-RNA-P also suppressed tumor growth in vivo, and combination treatment produced the strongest antitumor effect, reflected by reduced tumor volume and lower endpoint tumor burden. Together, these findings demonstrate that VACINUS-prioritized and peptide-validated neoantigens can be reformatted into a multi-epitope messenger RNA/lipid nanoparticle vaccine while retaining antigen-specific immunogenicity and antitumor activity. This study provides preclinical proof-of-concept for integrating AI-informed, TCR-aware neoantigen prioritization with messenger RNA/lipid nanoparticle delivery as a translational strategy for personalized cancer vaccine development.

immunology↗

bioRxiv2026

Pan-cancer single-cell analysis identifies a FOXF1/FOXF2-associated transitional CAF-like fibroblast state

Fibroblast heterogeneity shapes tumor progression, yet the transitional states linking normal-associated fibroblasts to cancer-associated fibroblasts (CAFs) remain poorly defined. Here, we integrated single-cell transcriptomic profiles of more than 90,000 stromal cells from 281 samples across nine cancer types to construct a pan-cancer atlas of fibroblast diversity. We identified a distinct CAF-like population positioned between normal-activated fibroblasts and established CAF subsets along the inferred fibroblast activation trajectory. Integration with single-nucleus chromatin accessibility data nominated FOXF1 and FOXF2 as candidate regulators of this CAF-like state. Functionally, CAF-like fibroblasts were characterized by non-canonical WNT activity, WNT5A-associated stromal communication, and a candidate GZMA-F2R/PAR immune-stromal signaling axis supported by spatial transcriptomics. Clinically, the CAF-like signature showed context-dependent prognostic relevance, with high expression associated with poorer survival in the tumor compartment of TCGA stomach adenocarcinoma. Together, this study defines a FOXF1/FOXF2-associated CAF-like transitional fibroblast state and links it to stromal signaling, immune-stromal communication, and cancer-type-specific clinical relevance.

cancer biology↗

bioRxiv2026

Molecular Alterations of Bovine Serum Albumin Induced by the Food Dye Acid Yellow 23: A Mechanistic Study

The widespread use of synthetic food dyes, such as Acid Yellow 23 (AY 23), in the food, cosmetics, and pharmaceutical industries raises questions about their potential effects on biological systems and public health. The concentration-dependent interaction between AY 23 and bovine serum albumin (BSA), a crucial model protein for understanding pharmacokinetics and protein-ligand behaviour, was examined in this study. We demonstrate that, under physiological conditions, increasing dye concentrations from 50 {micro}M to 200 {micro}M results in notable conformational changes, increased surface hydrophobicity, and protein aggregation using a multimodal biophysical approach that includes fluorescence spectroscopy. Direct visualisation verified these structural changes and aggregate formation, whereas hemolytic assay confirmed the high hemolytic nature of AY 23-induced fibrils. Additionally, this study provides a mechanistic basis for the toxicological effects of AY 23, underscoring the implications of food dyes for public health. HighlightsO_LIAcid Yellow 23 (AY 23) modulates Bovine Serum Albumin (BSA) structure and leads to aggregation under physiological conditions. C_LIO_LIStructural alteration is followed by binding of AY 23 at the hydrophobic regions of BSA, perturbing the globular protein into fibrillar aggregates. C_LIO_LI{middle dot} Confirmational changes induced by AY 23 in the BSA via interaction with Asp108, Pro110 and Ala193. C_LIO_LIFibrils formed after AY 23 interaction are observed to be hemolytic in nature. C_LIO_LIMolecular mechanism of AY 23-mediated fibrillation of BSA was assessed. C_LI Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=139 SRC="FIGDIR/small/737154v2_ufig1.gif" ALT="Figure 1"> View larger version (40K): org.highwire.dtl.DTLVardef@241c95org.highwire.dtl.DTLVardef@d0a03aorg.highwire.dtl.DTLVardef@c7445borg.highwire.dtl.DTLVardef@58cf7c_HPS_FORMAT_FIGEXP M_FIG C_FIG

molecular biology↗

bioRxiv2026

CDK1 facilitates RAD51-mediated DNA repair to protect dictyate stage arrested oocytes from genotoxic stress

Oocytes arrested at the dictyate stage of meiosis I must maintain genomic integrity for prolonged periods to preserve female fertility. During this extended arrest, DNA lesions arising from endogenous and exogenous sources threaten oocyte survival, yet the molecular mechanisms coordinating DNA repair in dormant oocytes remain poorly understood. Here, we identify cyclin-dependent kinase 1 (CDK1) as a critical regulator of the oocyte DNA damage response and homologous recombination (HR) repair under genotoxic stress. Using cisplatin-induced DNA damage models in fetal goat ovaries and neonatal mouse ovaries, we investigated repair mechanisms operating within the ovarian reserve. Label-free proteomic profiling revealed significant enrichment of DNA damage response pathways following cisplatin exposure, with CDK1 emerging as one of the most prominently upregulated kinases. Pharmacological inhibition of CDK1 had little effect on follicle survival under physiological conditions but aggravated oocyte and follicle loss following DNA damage, indicating a stress-dependent role for CDK1 in preserving ovarian follicle pool integrity. Mechanistically, DNA damage activated a Chk2-dependent signaling pathway that promoted p63 phosphorylation and altered the WEE1-CDK1 regulatory axis, resulting in reduced inhibitory CDK1 phosphorylation (Thr14/Tyr15) and increased activating phosphorylation (Thr161). Activated CDK1 was associated with enhanced RAD51 phosphorylation and accumulation at DNA damage foci, supporting homologous recombination (HR)-mediated repair in dictyate-arrested oocytes. In contrast, CDK1 inhibition reduced phospho-RAD51 levels, impaired RAD51 localization, increased persistent {gamma}H2AX accumulation, and elevated oocyte apoptosis. Notably, suppression of CDK1 was accompanied by increased expression of the non-homologous end joining (NHEJ) marker Ku80 and the nucleotide excision repair (NER) factor XPA, suggesting increased engagement of alternative DNA repair pathways. Furthermore, inhibition of Chk2 abolished the DNA damage-associated CDK1 activation signature and restored WEE1 expression, supporting a model in which CDK1 functions downstream of Chk2 signaling during the oocyte DNA damage response. Collectively, our findings identify a previously unrecognized Chk2-CDK1-RAD51 signaling axis that coordinates homologous recombination repair in dormant oocytes and safeguards ovarian follicular pool integrity under genotoxic stress. These findings provide new mechanistic insight into how dictyate-arrested oocytes maintain genome stability during prolonged meiotic arrest.

Cell Biology↗

bioRxiv2026

Additive baselines furnish no evidence for epistasis learning by MULTI-evolve

Recent work from Tran et al. (Science, 2026) introduced MULTI-evolve, a framework for protein engineering that combines single-mutant nomination via a protein language model (PLM) or a deep mutational scan (DMS), experimental single- and double-mutant characterization, and neural networks to engineer hyperactive multimutant proteins. The authors attribute the frameworks performance to "epistasis-aware modeling" and claim that their neural networks "learn the epistatic landscape" and "identify synergistic interactions" from limited double-mutant training data. Additive models, by definition, cannot represent epistasis, making them a natural null baseline for such claims. Here we show that MULTI-evolves multimutant predictions are almost perfectly correlated with an additive models across all three engineering applications (APEX, dCasRx, and HuABC2), such that the engineering of multimutants reduces to combining beneficial mutations with the largest additive effects--a standard protein engineering strategy for over four decades. We also find that MULTI-evolves neural networks do not outperform an additive model on held-out test set predictions, and do not even represent epistasis in their training data. Finally, we revisit a DMS benchmark finding presented as evidence of epistasis learning and show that the same pattern is expected even under a null additive model, due to an elementary statistical phenomenon; when we fit an additive model to the benchmark data, it reproduces the reported pattern. More broadly, our findings underscore the need to benchmark models for machine learning-guided directed evolution against additive null baselines before attributing performance to learned epistasis.

bioinformatics↗

bioRxiv2026

PARP1 directly disassembles nucleosomes to regulate DNA repair

Upon DNA damage, chromatin remodeling is rapidly initiated to promote chromatin accessibility, thereby facilitating the recruitment and assembly of repair factors. Although this enhanced accessibility has been linked to poly(ADP-ribose) polymerase (PARP) activity, the mechanism by which cells overcome the nucleosome barrier remains unclear. Using our designer chromatin system, we uncovered a previously uncharacterized activity of PARP1, whereby it directly and asymmetrically evicts histone dimers proximal to DNA strand breaks from nucleosomes to generate oriented hexasomes. In the presence of HPF1, PARP1 generates stable PARylated hexasomes, an open chromatin intermediate that can serve as a bifunctional hub for recruitment of DNA- and PAR-dependent factors. Using cellular assays, we demonstrated that PARP activity is both required and sufficient to drive chromatin accessibility and the recruitment of repair factors, with direct involvement of subnucleosomal species. Unexpectedly, we identified the C-terminal tail of histone H2A, a motif harboring recurrent cancer-associated mutations, as a critical determinant of efficient PARP1-mediated nucleosome disassembly. Deletion of the H2A tail sensitizes cells to DNA-damaging agents and PARP inhibitors, implicating a functional role of PARP1-mediated nucleosome disassembly in DNA repair. Together, our findings support a model in which PARP1 directly drives histone eviction, leading to the formation of subnucleosomes that facilitate efficient DNA repair.

molecular biology↗

bioRxiv2026

Ab initio prediction of RNA structure ensembles with RNAnneal

RNA utilizes three-dimensional structure in addition to sequence to carry out diverse functions in gene expression and disease. Much like well-folded proteins, RNAs adopt specific three-dimensional structures to carry out their function. Yet comparatively few RNA structures have been solved by atomic resolution structural techniques, in part because unlike structured proteins, RNAs fold into heterogeneous ensembles of interconverting structures that pose a challenge for high-resolution structure probing methods. In this work, we introduce RNAnneal as a method for RNA structural ensemble prediction that seamlessly integrates generative deep learning with statistical physics and molecular dynamics modeling. Given the primary sequence, RNAnneal uses ab inito (i.e., first principles) modeling to sample an ensemble of 3D structures, which, in turn, are used to train an ensemble of unsupervised deep learning models. The RNAnneal score, representing the consensus of the deep learning models, is then used to evaluate the 3D structures. We evaluated RNAnneal structures against 16 experimentally-resolved conformations (ERCs) of riboswitch RNAs and found that pseudoknot-free (PK-free) ERCs were well-reproduced by RNAnneal, with clear avenues for improving performance even on PK-comprising structures. Furthermore, we found that the RNAnneal score outperforms the Rosetta score and a state-of-the-art RNA forcefield on the task of classifying ERCs from decoys. We then introduce the interaction entropy as a measure of conformational heterogeneity within an ensemble and use it to assess our predictions. RNAnneal thus provides a generalizable framework for predicting RNA structural ensembles that will accelerate RNA-targeted drug discovery and the design of functional RNA molecules.

biophysics↗

bioRxiv2025

Insulin-Regulated Actin Dynamics is Disrupted in a Human Keratinocyte Model of Hailey Hailey Disease

Hailey Hailey Disease (HHD) is an autosomal dominant cutaneous disorder caused by mutations in ATP2C1, the gene encoding the Golgi/secretory pathway Ca2+-ATPase SPCA1. Characterized by suprabasal acantholysis and intertriginous blistering of the skin, HHD treatment focuses on managing symptoms as there is no cure. Challenges to targeted therapy are due to the lack of facile and reliable models, both human and rodent, for mechanistic studies. Here we validate and characterize CRISPR/Cas9 mediated single and bi-allelic ATP2C1 knockouts in immortalized human hTERT keratinocytes. Whereas SPCA1 expression, Golgi morphology and Golgi Ca2+ accumulation were proportionately affected in heterozygous and homozygous ATP2C1 null mutants as expected, both single and double allelic mutants showed near complete loss of cadherins associated with desmosomal and adherens junctions. HHD is characterized by poor wound healing and impaired keratinocyte migration. We show that SPCA1 is required for dynamic reorganization of actin cytoskeleton in keratinocyte spreading. We identified an insulin activated PI3K-AKT-Rac1 signaling pathway required for lamellipodia formation and keratinocyte spreading, defective in SPCA1 mutants. Transgenic expression of hSPCA1 or treatment with CDN1163, a small molecule Ca2+-ATPase agonist, restored defective phenotypes in the HHD model, paving the way for future therapeutic approaches to treat this disorder.

cell biology↗

bioRxiv2025

Lipid Droplet Remodeling Safeguards Redox Balance through the DGAT1-PANK2-NRF2 Axis in Mammalian Oocytes and Drives Age-Associated Decline

How lipid droplets (LDs) buffer metabolic stress and redox imbalance in aging oocytes remains poorly understood. Here, we identify de novo LD remodeling as a metabolic capacitor that couples lipid storage to mitochondrial fitness and oxidative resilience in mammalian oocytes. Live imaging revealed pronounced LD dynamics, with LD number peaking at metaphase I and declining by metaphase II, while LD area shifted inversely. Despite stable triacylglyceride and free fatty acid pools, {beta}-oxidation increased sharply, indicating elevated lipid turnover during meiotic progression. Spatial mapping and fatty-acid tracing demonstrated that newly synthesized lipids are actively incorporated into LDs, which arise primarily from the endoplasmic reticulum and engage with lysosomes and mitochondria. Acute inhibition of DGAT1, the rate-limiting enzyme of LD biogenesis, disrupted meiotic maturation and triggered oxidative stress, mitochondrial aggregation, and ultrastructural damage. Proteomic profiling revealed robust PANK2 upregulation and suppression of NRF2-linked antioxidant pathways. Mechanistic analyses showed that {beta}-oxidation blockade, PANK2 inhibition, antioxidant supplementation, or NRF2 activation each partially rescued DGAT1-dependent defects, and genetic validation in NRF2-null oocytes confirmed pathway dependence. Notably, aged oocytes exhibited reduced de novo LD biogenesis and impaired DGAT1-ER organization despite increased LD accumulation, resulting in smaller, metabolically inert droplets and a mismatch between lipid formation and utilization. Inhibiting PANK2 alleviated oxidative stress in aged oocytes, further implicating the DGAT1-PANK2-NRF2 axis in redox control and oocyte quality. Together, these findings establish LD biogenesis as a core metabolic capacitor safeguarding mitochondrial and organelle integrity during meiosis and reveal dysfunction of the DGAT1-PANK2-NRF2 axis as a mechanistic driver of reproductive aging.

cell biology↗

bioRxiv2025

Heterotic Pooling Strategy for Enhanced Hybrid Performance in Bitter Gourd (Momordica charantia L.)

Bitter gourd (Momordica charantia L.), a crop of significant nutritional and medicinal renown, is experiencing a surge in global demand, particularly in Asia. To meet this demand amidst the challenge of limited agricultural land, enhancing productivity through innovative breeding is essential. Hybrid breeding, which harnesses the power of heterosis, offers a proven strategy for boosting crop yields. However, the success of any heterosis breeding program, especially in a naturally cross-pollinated and genetically variable crop like bitter gourd, is fundamentally dependent on robust methods for assessing and strategically utilizing the available genetic diversity. This study compared the efficacy of traditional morphological trait-based clustering with a modern breeding value-based method (Heterotic grouping based on Specific and General Combining Ability - HSGCA) for predicting heterosis. Eleven diverse bitter gourd inbred lines were crossed in a half-diallel fashion to generate 55 F1 hybrids. The hybrids and parents were evaluated for key yield-related traits. K-means clustering based on morphological data resulted in five groups but proved unreliable for predicting heterosis, as high heterosis was often observed within clusters. In contrast, HSGCA-based clustering organized the lines into three coherent groups, where crosses between the two main groups (Group 1 and Group 2) consistently exhibited significantly higher mid-parent (20.89%) and better-parent (7.5%) heterosis compared to intra-group crosses. The HSGCA method provides a far more reliable framework for defining heterotic groups in bitter gourd, enabling more efficient and predictive hybrid breeding programs.

plant biology↗

bioRxiv2025

MutAIverse: An AI-Powered, Mechanism-backed Platform for Discovering Novel DNA Adducts and their precursor Genotoxins

Genotoxin exposure leads to DNA adduct formation, potentially causing mutations if unrepaired. Current DNA adductomics platforms or analytical workflows are limited by incomplete spectral libraries, reliance on experimentally validated adducts, limited cellular contexts, and inefficient computational methodologies. We introduce MutAIverse, an advanced AI-driven DNA adductomics analysis platform that overcomes these limitations by leveraging intracellular mechanistic modeling of genotoxin bioactivation to construct a comprehensive DNA adduct library and offers advanced spectral mapping. MutAIverse integrates experimentally validated and chemically valid putative DNA adducts, enabling interpretable retrograde tracking to parental genotoxins. Validation of MutAIverse against experimental MS/MS datasets demonstrated the detection of both known and novel adducts. Furthermore, application to in-house generated DNA adductomics data from tissue biopsies of smokeless tobacco-induced head and neck cancer patients revealed selective enrichment of both novel and known DNA adducts; a subset of them was validated using MS/MS analysis. Collectively, MutAIverse provides a robust, end-to-end, and interpretable platform for advanced DNA adductomics analysis.

bioinformatics↗

bioRxiv2025

Exploiting an Epigenetic Resistance Mechanism to PI3 Kinase Inhibition in Leukemic Stem Cells

Acquired non-genetic resistance mechanisms to existing therapies contribute to poor outcomes for acute myeloid leukemia (AML) patients, and inability to target leukemic stem cells (LSCs) can lead to relapse. To overcome these challenges, we tested whether LSCs have dependencies on PI3 kinase (PI3K). We found that LSCs are susceptible to isoform-selective targeting of PI3K and are particularly dependent on the P110 alpha isoform of PI3K. We discovered that PI3K inactivation leads to dynamic changes in EZH2/PRC2 function in leukemic cells, and we uncovered downregulation of EZH2 protein levels as a resistance mechanism in response to PI3K inhibition. We found that PI3K inhibition in AML cells can lead to compensatory upregulation of EZH1, and that EZH1 knockdown can sensitize AML cells to PI3K inhibition. We leveraged this resistance mechanism by combining a PI3K inhibitor with an EZH1/2 dual inhibitor, which successfully overcomes the acquired resistance and leads to sustained targeting of AML cells ex vivo and in murine AML and PDX models in vivo. This study identifies a promising novel therapeutic regimen for targeting LSCs in AML.

cancer biology↗